Scholarlink

<p>Falling detection among elderly individuals is a critical research domain for enhancing safety in ageing populations. While existing surveys predominantly focus on machine learning (ML)-based algorithms, this comprehensive review breaks new ground by systematically analyzing a broader spectrum of algorithmic approaches, including sensor fusion techniques, rule-based systems, and hybrid models beyond the narrow scope of ML. We curate and categorize studies (2018–2025) from major databases, examining tools, sensors, and temporal dataset trends. Our analysis reveals key advancements in both wearable and non-wearable systems, highlighting emerging trends that were previously overlooked in prior reviews, such as the rapid adoption of multi-sensor fusion in non-wearable systems (e.g., radar + RGB-D in 68% of recent studies) and the increasing use of explainable, rule-based algorithms for clinical interpretability. This synthesis not only maps the current landscape but also identifies understudied niches for future research, particularly in privacy-preserving edge computing for camera-based systems and validation on cohorts with neurological disorders.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Comparison Framework to Survey Elder Falling Detection Based on Machine Learning and Sensors Used Approaches

  • 期刊论文

  • 发表日期：

    2026年4月29日

查看全文
  • Samireh Mahmdnezad,
  • Amin Babazadeh Sangar,
  • Mohammad Masdari,
  • Kambiz Majidzadeh

摘要

Falling detection among elderly individuals is a critical research domain for enhancing safety in ageing populations. While existing surveys predominantly focus on machine learning (ML)-based algorithms, this comprehensive review breaks new ground by systematically analyzing a broader spectrum of algorithmic approaches, including sensor fusion techniques, rule-based systems, and hybrid models beyond the narrow scope of ML. We curate and categorize studies (2018–2025) from major databases, examining tools, sensors, and temporal dataset trends. Our analysis reveals key advancements in both wearable and non-wearable systems, highlighting emerging trends that were previously overlooked in prior reviews, such as the rapid adoption of multi-sensor fusion in non-wearable systems (e.g., radar + RGB-D in 68% of recent studies) and the increasing use of explainable, rule-based algorithms for clinical interpretability. This synthesis not only maps the current landscape but also identifies understudied niches for future research, particularly in privacy-preserving edge computing for camera-based systems and validation on cohorts with neurological disorders.